Avatar

Shreyansh Singh

Lead Machine Learning Engineer

Level AI

About Me

I am currently working as a Lead Machine Learning Engineer at LevelAI. My team is is spread across India, USA & Canada and we primarily work on solving challenging NLP problems at scale to extract and deliver insights from contact-center conversations. Previously, I was working at Mastercard’s AI Garage as a Data Scientist where we were leveraging AI to make the transactions world a smarter and secure place. In my corporate experience, I have mainly worked on applied research, meaning I have been involved both in incorporating AI and Machine Learning in product development (leading to product launches/patents) as well as pure research work (publishing papers at AI conferences).

I did my B.Tech in Computer Science and Engineering from IIT (BHU) Varanasi. My B.Tech thesis advisor was Prof. K.K. Shukla. I worked on privacy-preserving Machine Learning and its application in the medical industry. Previously I had also worked with him on Adversarial Machine Learning and Malware classification problems. In summer 2019, I was an intern at Samsung Research Institute - Bangalore, working on mobility in 5G networks. I also worked as a research intern at the C3i Institute, Indian Institute of Technology Kanpur on Malware Detection for Linux, in winter 2018. In summer 2018, I was a Data Science intern at Innoplexus, Pune (India), working with the Computer Vision team. I have also worked with Dr. Anil Kumar Singh, on sentiment analysis on product reviews and natural language generation through shared tasks at conferences.

When not working on research, I enjoy playing CTFs (wr47h), either solo or with my team Abs0lut3Pwn4g3. I also enjoy competitive programming and reading novels.

Interests

  • Machine Learning
  • Natural Language Processing
  • Computer Vision
  • Cryptography

Education

  • B.Tech in Computer Science, 2016 - 2020

    Indian Institute of Technology (BHU), Varanasi

Experience

 
 
 
 
 

Lead Machine Learning Engineer

Level AI

May 2023 – Present (Remote) Mountain View, California
 
 
 
 
 

Senior Machine Learning Engineer

Level AI

Sep 2022 – May 2023 (Remote) Mountain View, California
 
 
 
 
 

Machine Learning Engineer

Level AI

Jan 2022 – Sep 2022 (Remote) Mountain View, California
 
 
 
 
 

Data Scientist

AI Garage - Mastercard

Aug 2020 – Jan 2022 Gurgaon, India
 
 
 
 
 

Student Trainee

Samsung Research Institute - Bangalore

May 2019 – Jul 2019 Bangalore, India
Worked on mobility in 5G
 
 
 
 
 

Research Intern

C3i Institute, IIT Kanpur

Dec 2018 – Jan 2019 Kanpur, India
Worked on Linux malware detection
 
 
 
 
 

Data Science Intern

Innoplexus AG

May 2018 – Jul 2018 Pune, India
Worked on automated text and table extraction from academic papers (PDFs)

Latest

Promoted!

Promoted to Lead ML Engineer!

Promoted!

Promoted to Senior ML Engineer!

New job! New challenges!

Joined LevelAI as an AI Software Engineer in NLP.

Paper accepted

Our paper ‘MeTGAN: Memory efficient Tabular GAN for high cardinality categorical datasets’ got accepted at ICONIP 2021.

Paper accepted

Our paper ‘CuRL: Coupled Representation Learning of Cards and Merchants to Detect Transaction Frauds’ got accepted at ICANN 2021.

Silver medal - Shopee - Price Match Guarantee Competition

Earned a Silver Medal (top 5%) in the Kaggle Shopee - Price Match Guarantee competition. Ranked 115th among 2426 teams while participating solo in the competition.

First job

Got my first job and started my career in the field of AI with Mastercard.

TAship for Artificial Intelligence course

Started to work as a TA for the Artificial Intelligence course offered to sophomores of the CSE Department of IIT(BHU) Varanasi.

Student scholarship

Won a student scholarship to attend BlackHat Asia 2019, Singapore. 100 students were selected from 82 countries.

Paper accepted

Our paper ‘IIT (BHU) Varanasi at MSR-SRST 2018: A Language Model Based Approach for Natural Language Generation’ got accepted in the Proceedings of the 1st Workshop on Multilingual Surface Realisation (MSR), 56th Annual Meeting of the Association for Computational Linguistics (ACL), July 2018, Melbourne, Australia.

Projects

*

Optimized NN Inference using custom Triton kernels

Implemented a high-performance linear layer (both forward and backward pass) with (optional) activation layer fusion using OpenAI’s …

Flash Attention in Pytorch

A simplified implementation of FlashAttention in PyTorch. I have implemented the forward pass and backward pass algorithms from the …

KV Cache in Nanogpt

A modification of nanoGPT to use KV-Cache during inference. Using a KV Cache helps speed up inference since we don’t have to do …

ConvNeXt - Adversarial images generation

I implemented Stanislav Fort’s project in Pytorch. The Github repo has a notebook which looks at generating adversarial images to …

ML Optimizers in JAX

Implementations of some popular optimizers from scratch for a simple model i.e., Linear Regression on a dataset of 5 features. The goal …

Privacy-preserving Deep Learning for Medical Image Classification

Perform medical image classification in a secure and privacy-preserving manner using Secure Multiparty Computation and Differential …

Network Intrusion Detection in an Adversarial setting

A study on fooling Machine Learning/Deep Learning based Network Intrusion Detection systems to prevent them from detecting intrusions

Linux Malware detection using Machine Learning

Implemented various papers on Linux Malware detection, where I analysed the structure of ELF files to determine whether they were …

Multilingual Surface Realization for NLG

Our system for a Narural Language Generation based shared task organized at ACL 2018 (Association for Computational Linguistics, …

Review Opinion Diversificatio­n

Baseline model for RevOpiD, a shared task organized at IJCNLP 2017 (International Joint Conference on Natural Language Processing, …

Worldlink - Social Networking Website

A social networking website made using Django

Publications

MeTGAN: Memory efficient Tabular GAN for high cardinality categorical datasets

Generative Adversarial Networks (GANs) have seen their use for generating synthetic data expand, from unstructured data like images to …

IIT (BHU) Varanasi at MSR-SRST 2018: A Language Model Based Approach for Natural Language Generation

This paper describes our submission system for the Shallow Track of Surface Realization Shared Task 2018 (SRST′18). The task was to …